AI Adoption in Mid-Market Companies

Welcome to the Digital Transformation Podcast, interviews with best-selling authors, innovative thought leaders, and top shelf executives all driving today's digital success. This is the show that will help you take advantage of digital transformation to build your business and career. I'm your host, Kevin Crane, and I'm so pleased that you're listening. Our guest today is Justin Watt, Justin is the CEO and co-founder of Switchboard, where he helps mid-market companies replace manual workflows with automated scalable systems. With deep expertise in operations and AI adoption, Justin is all about bridging the gap between individual productivity tools and team-wide systems that actually work in practice, and he is with us today to discuss why AI adoption often fails, and how companies can build operational systems that deliver at scale. So, Justin, welcome to the program. Indeed, analyst firms like McKinsey, Bain, and others all point to research that indicates that AI initiatives have a high rate of failure, often 70% or more, or many companies invest in AI, why do so many AI projects struggle to take hold inside real organizations? Yeah, I think it's fascinating to watch the excitement around technology advancing, but the challenges where most things fall apart of the test of time it seems. And we've seen it ourselves with folks who come to us after working with some firms where it's really the change management. It's the people side where it falls apart or fails to gain traction. I always, whenever folks bring that challenge to me, we're scared to try because we think it'll fail, or we have tried it and it already failed. How should we go about this differently? The conversation often turns into, well, you know, would you hand a 16-year-old with their learners, permit keys to the car and say, good luck? Or would you probably hop in there and say, okay, this is how you should check in parallel park, and this is when you should use your signal. And that car is frankly dangerous, just like a lot of this technology is if it's, you know, not used properly, anything about security permissions. But that's kind of the size of the point. The main thing is this is affecting how humans work and how they go about their work. And a lot of it is, you know, let's pilot something, give the kid the keys to the car and hope they figure it out and it works out well for us. And that seems to be when things fall apart. Now let's talk a little bit about single player AI versus multiplayer AI. You draw the distinction between single player AI, which helps individuals and multiplayer AI, which operates across teams and workflows. How does that difference affect adoption and ultimately outcomes? Yeah, I think a lot of people are really excited by their individual use case for AI. And so what that looks like for most people is I've opted to chat to the team play around with it or quad or quad code. And if they so desire to go that far down their rabbit hole personally to try out AI, but they often forget how does this apply to a team, the 10 minutes I spend, you know, giving context and giving access to tools and all the different things that go into getting your individual needs out of the I are very different than if you're trying to quote unquote inject AI into the entire system and workflow of business. So often something like let's call it new employee on boarding that employee on boarding doesn't just touch one person that is usually has four or five six people in the mix between managers and I team whatnot. And so for these systems to help out with people's work, it's very different kind of like video game that single player experience is very different than if you're playing online, you know, with a bunch of friends, you have headsets and you're talking with each other. And I think a lot of folks are eager to start, but not always set up to think about their business differently when it's call it multiplayer AI instead of that single player experience. So what should we be doing about that? I you know, if I'm employing AI across my organization, I want multiplayer AI to operate across teams and workflows to bring them more efficiency and efficacy. What's one thing I can do to make sure that that happens? Yeah, I think a lot of folks start with, well, if it works for me to work for everyone, but there's a very large difference between the data that's used across an organization and that which is used for an individual. So I always tell folks to start with first thinking about how does data flow through our organization and the absence of technology and humans and process. What is the data point? So again, taking an example to say a new customer client or user. What is the entry point for them into your systems of record? Where does that data go to? Who else uses it? Because again, kind of going back to that multiplayer. If someone cusser support doesn't have an AI that has access to that same system that someone in sales had or someone in operations has, then your AI starts to fall apart pretty quickly if that data isn't speaking to each other, so to speak. So that's often where I see teams not thinking about how they need to adapt their data first before they actually can inject this everywhere, if that makes sense. There is a concern or a great deal of discussion about moving from pilot initiatives to a broader production application across a company or an enterprise. So when implementing AI and automation, how should companies balance the need for quick wins with building long term scalable processes that actually stick? Yeah, I someone said this the other day when we were on a call like do you want to sprint AI or do you want to have a marathon on the way there? And really then underlying question turn and discussion turn into do you want this to permeate everywhere and you want to set up the business for that from day one, which goes back to our discussion on data and how we figured out a process. Or are you looking for that quick hit of dopamine to say this one process is so broken, how can we get AI and they're very quickly. And then you have to worry less about company wide data and process. So usually the rapidity of ROI of this stuff looking at through business lens is where I encourage people to start if you understand inherently that it Rome was not built in a day and you want AI everywhere in the business, then that's a very different starting point. Then we're going to apply it to this one workflow, usually with folks, though, dipping their toes in the water instead of not wanting to jump into the deep end. I always say folks on one area of the business first and test out a small pilot because a lot of times I think we hear about pilots failing, but a lot of times it turns out that pilot was well, we tried to pilot everything all at once and it fell apart. We don't know why. So I guess re-vraming of the word pilot to make sure that folks understand, you know, when they're thinking about how they want to apply this, what is the adventure I want to go on and how long do I want it to be and then have we pilot it with a true pilot, which in my opinion would be a small group. One workflow, one area of the business, that kind of thing. Do you want to be a guest on the Digital Transformation Podcast? Well, it could happen. Tell us about your ideas and advice. What technologies and solutions do you recommend and how can our listeners benefit from your approach? Be a guest. Find out more at digitaltransformationpodcast.net/guest. You're listening to the Digital Transformation Podcast. We are here today with Justin Watt. Justin is the CEO and co-founder of Switchboard and he's all about bridging the gap between individual productivity tools and team-wide systems that really make a difference. You can find out more and find Justin at withswitchboard.com. Now Justin, many organizations are still working with spreadsheet-driven processes that are common, especially in mid-market companies. What are the biggest challenges when moving from manual workflows to automated team-based systems? I find a lot of folks just get caught up conceptually. They can't imagine that world in which they don't have to open six spreadsheets, enter in a bunch of information, send three of those out by email, ping another three people on Microsoft Teams to say, "Hey, this is up to date and you can do your part of the process," or whatever it might be. Again, those pilots being very specific and very small just to show that to the rest of the organization. Then people I find it's kind of striking gold. Once you hit it in one area, you're looking for it everywhere and you want the map to it. Until you see that first piece of gold, you're kind of like, "I don't conceptually like where would I find this? What would I do with it?" Thinking of things with that lens, and then also just reminding folks I find a lot of times leadership to remind folks, Excel came out the same year that back to the future came out into theaters. The first back to the future movie. This is a tool that is 40 years old. A lot of people, it's just all that they know. Getting people in that mindset again, the world has adapted and evolved since back to the future came out, even though that was set in the future. Point being that is a very long time for people to get accustomed to something. Leadership setting that tone of that does work for you, but that is not how a modern business operates. There's no reason that six spreadsheets have to be updated to unlock two other teams to do another part of the process when you're working in these mid-market companies at scale, getting caught up. This is how we've always done things with spreadsheets as kind of half the battle with folks. Well, it's sort of a matter of trust. I mean, I'm going to trust a spreadsheet, 40 years old, proven technology. I may not trust new systems in terms of data sharing, security, AI. Do you think that's changing, however, though, as leadership and workers all across the board are really a new generation or even a second generation past back to the future days? Yeah. I think that's why I always suggest folks just start playing around with AI at a personal level. We're trying that solo versus multiplayer approach, but leaders understanding the progression of trust that you can put in it. And I agree with that sentiment too. Like I'm rarely the first to say, yes, your pilot should be this huge undertaking that takes six months because you're testing it in every department with all your data. There's very real risks if you don't do it properly. There's huge benefits, but great risk, great reward kind of thing. People have to think about this security permissions. Someone can't type into a company wide chat bot. How is so-and-so doing in their performance review? If they should not be able to see that in the system, they should not be able to ask an AI for that as a simple example. So thinking about that side of it, but also the team's really understanding that the technology's evolving quickly. That's the change I'm starting to see. Because again, that's why I'm often the first to say, let's start with a small pilot and area of focus. Because the technology is, people don't hire us, for example, thinking they can go work on a beach in three weeks. The technology is not there yet to trust it. So again, I think it's folks at a leadership level trying. So then they can kind of see the guard rails wake and trust it where you want as the term often comes up, a human in the loop. So it's something that given that there's millions of use cases, you want to be sure that it fits yours. You've mentioned starting with a small pilot first. And of course, that makes sense. The trick is now moving from that small pilot stage. Now, what happens? What can I do after I've done a pilot? It's been successful. The trick is moving to a broader application. What are some techniques that I can use after that successful pilot to engender the sorts of support and acceptance that I need to move to a broader in production phase? Yeah, I think that's where taking the pilot and showing it to the rest of the team together. Their heads wrapped around conceptually how this stuff can work is really valuable, I find. So a lot of times the folks that are looking for a pilot are looking to solve an immediate challenge. The other great benefit of that I find is that you can then show it to the rest of the org. So as a quick example, there was a team. They were about 120 person architecture firm that we were working with in the whole org one-a-day eye, but that it stopped there. That's all they said. And so the pilot that we chose was as an architecture firm, they are constantly looking for RFPs for new jobs to bid on. That was three or four people spending about 10 hours a week doing that. So we built an agentic system that goes to all the municipal websites and all of the universities and whatnot to find these RFPs. Bring them in, scan them to see if they're fit for this team based on their criteria. All these elements that are low risk as a pilot and small and focused, but let them see how they're thinking and their internal context can be applied to saving them, you know, combined 40 hours a week on their sales team of searching for RFPs. That specific example is very specific to them, but showing the rest of the org that it clicked for, you know, half a dozen other departments to go, oh, I see how that can apply to what I'm doing here. And so that's again where I think that pilot to production approach is folks starting small with the data and growing and then also starting small with the thinking and showing that to people. So then that grows as well. That's a great example. What happened to those RFP teams afterwards? Well, that's that's the thing that I love seeing. Like I know that there's a lot of fear around AI, but there's a, in my experience, there's always going to be more work to do. Businesses have ambitions. The people that run them have ambitions. The people in their roles, but their careers have ambitions. And so I've yet to see, you know, a case where AI is brought into a business and there's now a mass reduction in headcount. It's this freeze up time for more important things. I think we all grew up with the dream of astronaut Dr. Lohler, no one grew up with the dream of moving Excel spread sheets around all day kind of things. So I've seen so far that that team is an example. That was about 10 hours a week, ish that four or five people got back. And so they are now doing more customer relationship building and spending more time on outreach and whatnot and crafting, you know, better proposals so that they do with these beds that are found for them. So I found yeah, it tends to bring a lot more time for more important things for people. Now you describe operations as a product that evolves as company scale. What do you mean by that? And what does it mean to treat operations this way? And how can leaders start doing it? Right. Yeah. I'm biased because I before kind of co-founding switchboard came from working in, I guess with tech startup land and Silicon LA land where everything is a product and by product people mean an app or a service that evolves on that some sort of digital software. And so I found though that that thinking applies to business and it clicks for a lot of leaders who have a hard time thinking about how do we evolve as a business in this new era is, you know, the version of Uber in the app store is not the same as Uber five years ago. They've added features. They've changed the layout. They changed out things work. They've evolved the actual look and feel of it. All of that stuff comes together to equate to a business in terms of the business that you were trying to run when there was 10 people is different than 100, which is different than 1000. And having that kind of product thinking of what is the release version, you know, when you go into an app store and you hit update, sometimes that brings new features, sometimes it brings fixes. And so instead of treating it as this constant sprint to figure out how the business should operate, treating as more of that, what is the product, what is the kind of release stage is what features are in that release really helps people think about their business differently. And I find helps us because technology tends to be released iteratively so we can do it alongside business change. Is it time to reach a new audience in a new way? Advertising on the Digital Transformation podcast gives you the opportunity to do just that. Each week you'll reach thousands of listeners all tuned in to learn about strategies, products and approaches that will help them succeed. Be a sponsor and get your message heard by the right audience. Learn more at digitaltransformationpodcast.net/advertise. That's digitaltransformationpodcast.net/advertise. We are speaking today with Justin Watt from Switchboard with deep experience and operations and AI adoption. Justin is helping companies in mid-markets bridge the gap between individual productivity tools and team-wide systems that really move the needle when it comes to digital transformation and AI. Now Justin, moving from pilot to production in AI adoption, it seems to me that a critical need is to demonstrate the success of that pilot and how it may scale up. Can you help us understand what metrics or indicators companies should track to measure whether automating operations is actually delivering value and improving scalability? Yeah, I would say there's the technical side of that and the people side of that. Both of them tend to equate to at the highest level a better margin in the business because you should be able to do more as a business with less time spent on administrative things. So from that lens, the two things I always bring up to folks is on the technical side. If folks aren't familiar with this term, look up eVALs, evaluations in AI. So that is essentially the effectiveness of the AI system that you have set up. So in this era, there's quickly tools on the technical level around being able to evaluate if something that someone did a hundred times a week is now done by AI, how accurate is that hundred times the AI is doing it? So then you can see the measurement. It's a real quantifiable measurement for folks to say, is this effective? Is this working or not? So that's more on the technical side. On the people side, I was the way we work with folks because it clicks for them quicker. And what I always recommend for anyone wanting to do it is measure out the process that you are trying to automate or add AI to. And a lot of times that practice of the process mapping helps in of itself to clarify where AI can help. But more importantly, as you're mapping it out, mark down how often do we do this and how long does it take human to do? And then after roll out, you then have a baseline to say, okay, it's a week, it's a month, it's a quarter later, we're going to go back to those seen people and go through the process and say, how much time are you spending on this step now? Where AI was supposed to help or automation was supposed to help. And then you very quickly again, got a quantifiable number. And then the last thing on the people's side is asking the people their sentiment towards it that this helped at free up time. That's less quantifiable, but the human part of this is important too. So those are kind of the three areas. Now, Justin, you are a specialist in mid-market companies. Now, we talk about AI adoption. Oftentimes it's with respect to larger enterprises that have teams of experts and rather large budgets. But for mid-market companies with limited resources, where do you think that they should start when thinking about AI adoption and really integrating it into their company in ways that make a difference? Yeah, we like to work with mid-market companies because they have that, I guess, call it operational maturity. They likely have some people at the table already internally who can make the stuff happen. But the first question for a company of that size is usually what people do we have internally versus what we need to make happen because AI and automation is very kind of too passive expertise. There's the business side and the technical side. And a lot of that's very hard to find. That's why a lot of folks hire us. But this isn't about pitching what we do. It's about saying that internally a lot of folks will try to start a pilot saying, "Oh, we have our COO and his team loves process and they'll figure this out." But do they have the right technical person on that team to think about what we've tried about, everything from e-thals and what's the ROI to their really stages of what is our data like and what is integration technically look like for this. And so if a team doesn't have that, both of those skills internally, whether that's two separate people or one who has it a whole, is really like taking that step back and self-assessing, do we have the right people internally to help out with this? Because if you don't have that blending of the technical, very heavy technical understanding and the heavy business need and solution understanding, that's where it tends to go back to your earlier point of pilots fall part, adoption efforts fall part of that kind of thing. Can you give us an example of one organization or one use case that you feel has been particularly successful? What did they do and what were the results? Yeah, we had a team that we actually worked with fairly recently where they, like every business, the first discussion was, well, we're in a niche and it's hard to understand our industry and whatever that might be. Most businesses at the end of the day, if you take away all of the what's your pipeline, what's your delivery, is that a user product customer, it tends to be data moving through a business of, well, that customer started as a lead and went into a CRM and then went over here. And the teams, the few that we've seen where we're there is more of an advisor than having to get super into the weeds on everything, is those who understand their industry deeply, because that's usually where the opportunity for automation is because there's a lot of off-the-shelf tools that teams will use to replace spreadsheets. The whole industry of SaaS is about buying tools off the shelf that can help you move your workflows quicker, but that there's so many businesses where SaaS can help with a little bit, but what can't it help with? Because again, kind of back to your point of this being focused on mid-market companies, they don't have the enterprise budgets of a legion of software developers to build custom stuff for them everywhere across the business. So those who have found it be successful have really thought about what can we buy off the shelf and then what do we need custom and what will we need that support on? Because usually it becomes very easy conversation for we already know the value, we know the need, etc. And so that's one way that, you know, businesses that are successful with this, they start with that thinking first. Well, Justin, we have reached the action item round of the program. I'm wondering if you could please provide us with three quick action items that our listeners can use to take advantage of your ideas and advice. Yeah, I think I won't treat this as a summary of everything we just discussed, but I'm hesitant because I would love to kind of recap. Like the big thing is to understand what a pilot is for your team and really figure out and get granular about that to make sure it's it's shaped enough where it yes, it can help as our why, but a pilot does not meant to be the one one stop of AI. It's meant to be a starting point. And so I'd really encourage folks as an action item. If they're thinking about what does AI look like in my business, it's not let themselves get overwhelmed by opportunity or fear of where to even start and getting it wrong. That's why I would say really think about your pilot first. The other thing that I found is as an action item is try to map out some of your process. A lot of, you know, what we do is saying if a step, if a process in a business has 50 steps, regardless of technology, like just sitting down and saying, we're going to reinvent how we work any ways, how can we reinvent regardless of technology? Could that 50 steps become 30 and of those 30, then which ones do we automate in that AI too? That is a benefit when you get today, I implementation phase of already having your process mapped out, but I think a lot of folks just get overwhelmed by the opportunity or overwhelmed of it all. And so just starting with mapping a few of your processes and that can be blocks on a screen or whiteboard. It doesn't have to be super fancy, you know, 20 page documents. It can just be we're just going to map out what happens when and who does it. And then the final thing, probably the most basic, and I know others have said it before here as well, but just starting if you haven't like again that solo, solo multiplayer conversation that we had. A lot of these folks are scared to even start with the solo part of opening up chatgvt or cloud.com and just creating an account and trying it. You don't have to give it access to your tools. You don't have to throw personal or company data in there to be able to understand quickly how it can work for you as a business. So just starting, I know that it sounds silly to say because a lot of folks listen to this or like, no, I'm doing all that, but there's billions of people who haven't even tried that part yet. So that's where I would start. Well, Justin, it has been great speaking with you today. We're almost out of time, but before I let you go one last question, what should executives and business owners be thinking about now? And strategizing for today in order to be prepared for the world in five years time. The five year question is mine's much greater than mine or are also pontificating on that, but I think the biggest thing is this is not going away. This is, you know, I think the bubble of the internet era that a lot of people compare to is we will invest in building the infrastructure and hope that the people will come and AI so far has been the opposite of they cannot keep up enough all this infrastructure is being built and all of it is prepaid for years in advance because the the frontier model companies and the large call it hyperscalers, they know where this is going. And I think that's the biggest thing is leaders need to understand that this is not a bubble and there is way too much hype and snake oil, I agree, but this is going to fundamentally, it already is, but it will continue to sort of play out your five year question. It will fundamentally change every business and how they operate or they will not exist anymore, not because of some fearmonger, you know, capitalistic statement. More so as it's going to become very obvious to many people who are consumers or B2B buyers that I'm going to go with this company because they're using AI properly, whether that's a product I'm using or how I interact with them and I'm saving money, I'm working with these folks and they probably do it better and faster. And so it's just going to naturally evolve where over the next five years, this has to be adopted by folks. So that's how I would think about it. That is Justin Wat with Switchboard. Justin, it's been great speaking with you today. Thank you so much for being our guest today on the Digital Transformation Podcast. Yeah, thanks Kevin, it was great to be here. That'll do it for this episode of the Digital Transformation Podcast. But join me next time when I continue to talk to best-selling authors, innovative thought leaders and top-shelf executives, all driving today's digital success. And I'll talk to you next time on the Digital Transformation Podcast. Do you want to be a guest? Marketing is hard. But I'll tell you a little secret. It doesn't have to be. Let me point something out. You're listening to a podcast right now and it's great. You love the host. You seek it out and download it. You listen to it while driving, working out, cooking, even going to the bathroom. Podcasts are a pretty close companion. And this is a podcast ad. Did I get your attention? You can reach great listeners like yourself with podcast advertising from Lib Sin ads. Choose from hundreds of top podcasts offering host endorsements or run a pre-produced ad like this one. Across thousands of shows to reach your target audience and their favorite podcasts with Lib Sin ads. Go to LibSyn ads.com. That's L-I-B-S-Y-N. Ads.com today.